We will start off this week with a nota bene

There will be an actual scouting report at the end of this column.[i]

While plugging through Top 101 write-ups around Christmastime, usually with an old fashioned[ii] nearby to try and fire the synapses labeled “inspiration,” I slowly became aware of two things.

1. Shortstops, so many shortstops

2. We need to contrive more synonyms for “mid-rotation starter”

All in all, Wilson, Adam, and I wrote up 14 players that could be described as having a mid-rotation projection. That might even be low. I got so tired of writing the same entry over and over that I even indulged in a bit of Mad Libs for poor Braden Shipley:

This was still on my mind a few months later when I came up with a potentially harebrained theory: There is no such thing as a mid-rotation starter projection. It’s the Dunning-Krueger effect transposed to pitcher evaluation. We don’t know that we don’t know.

I had been thinking about Rafael Montero, the kind of “mid-rotation” prospect that shows up towards the back of national lists.[iii] He didn’t fit neatly into the mad libs above, his command was above-average, his best secondary pitch was fringy, but at that point you are just sliding some nouns around anyway.

I saw Montero in Double-A, and he pitched quite well. That was always going to be the test for the profile, right: A short righty without a huge fastball that gets by on command? He dominated the Eastern League as well. In 66.2 innings in Binghamton in 2013, he struck out 72 and walked just 10. While the stuff didn’t jump off the page, it was good enough, and you could talk yourself into an average slider and change. With the strong command profile that is a future no. 3, likely no. 4. It’s almost pro forma.

It didn’t quite work out that way. Montero has had a few injury issues, most notably a mysterious shoulder ailment that kept him on the shelf for most of 2015. After talking him up a ton in Spring in 2014, the organization seems to have soured on him, and the plus command hasn’t shown up at the highest level.[iv] The secondaries didn’t make it to average either. Now Montero looks more like a swingman or middle relief type, and one that may need a change of scenery to even achieve that role.

Of course, that is one example, an anecdote.

But the plural of anecdote is data, right?

For a slightly larger case study, I went back to the 2012 Baseball Prospectus Annual to take a look at pitchers who made that Top 101 while being described as a mid-rotation type. To be fair to Mr. Goldstein, I tried to interpret this as literally as possible, though KG was better at contriving synonyms than I am.

#34 Danny Hultzen: “more of a good, consistent pitcher than a star.”

Injury attrition

#41 Randall Delgado: “His stuff is that of a mid-rotation starter”

Below-average starter that got moved to the bullpen. Was a mediocre reliever in 2014, a good one in 2015.

#50 Jarrod Parker: “and year two back from the procedure will give Oakland a much better feel whether he’s a future number two or number three.”

Fits the bill as a mid-rotation type before a catastrophic series of injuries

#53 Robbie Erlin: “while his ceiling is a bit lacking his floor is as high as nearly anyone’s.”

Up-and-down sixth starter/swingman type.

#64 Brad Peacock: “His ceiling is a number-three starter, but he’s arguably that right now.”

Up-and-down sixth starter/swingman type with home run issues.

#74 Joe Wieland: “and both [Wieland and Erlin] are going to be solid big league starters.”

Emergency starter.

#78 Casey Kelly: “he still looks like a good future number three.”

Up-and-down sixth starter/swingman type with injury issues.

#82 John Lamb: “[he] could turn into a left-handed version of Jeremy Hellickson.[v]

Still probably too early to make a definitive call, but early returns haven’t been great.

This isn’t intended to impugn KG. Plenty of stuff that I wrote in the 2016 annual will look far worse than the above by 2020. One could easily invoke the scouting shibboleth, TINSTAAPP, here as well. And you will miss on all kinds of arms. The aces bust and the future pen arms over-perform. But I think the mid-rotation grade is a specific sort of hedge.

We know what top-of-the rotation stuff looks like. Those guys are easy scouts for the most part[vi]. Sure, it doesn’t always work out, but when it doesn’t the culprit is fairly obvious (and usually injury-related).

And we know what back-end dudes looks like. We see a lot of them. Our reports are peppered with the word “fringy.” Their best pitch is probably a changeup. You’ll miss here too, but no one really cares if you don’t actualize a role 45 arm. It’s actually a player development win, mind you, but for most fans, you’ve just been told: “Dude, you’re getting a Dell.[vii]

As for the mid-rotation arms…

Okay, some actual data

So what does a major league no. 3 look like? I had the stats team compile a three-year sample of DRA and include standard deviations. Now, the scouting scale is based on the bell curve in theory, but major-league performance doesn’t neatly fit along a parabola. The farther left you go on the curve, the fewer opportunities are given to players. This is as you’d expect. There is a much larger population of well-below-average major leaguers than the number that get actual major-league time.[viii] From 2013-15, only one pitcher threw more than 300 innings as a starter while being more than one standard deviation worse than the league by DRA.[ix]

At the other end of the scale, every pitcher one SD better than league average you would comfortably call a top-of-the rotation starter. The list:


Clayton Kershaw

Jake Arrieta


Max Scherzer

Johnny Cueto

Sonny Gray


Zack Greinke

Matt Harvey

Chris Sale

Adam Wainwright

Jacob deGrom

Yu Darvish

David Price

Corey Kluber

Chris Archer

If we are strictly adhering to normal distribution, our role 6 (no. 3 starter) would be around one standard deviation better than the league, and the next tier down feels like it has fewer top-of-the-rotation names in it.


Anibal Sanchez (0.999, almost feels unfair not to round, but also kind of ruins my neat dividing line)

Alex Cobb

Carlos Carrasco

Hisashi Iwakuma

Cole Hamels

Garrett Richards

Cliff Lee

Collin McHugh

Dallas Keuchel

Hiroki Kuroda

Madison Bumgarner

Justin Verlander

Clay Buchholz

Michael Wacha

Tyson Ross

Stephen Strasburg

Gerrit Cole

Shelby Miller

Jon Lester

Jorge de la Rosa

Francisco Liriano

Danny Salazar

Hyun-Jin Ryu

We are a bit at the mercy of a three-year sample size here. Guys get propped up by one good year, or dragged down by one bad year. We also have huge inning ranges, from 304 (Cliff Lee) to Madison Bumgarner (636). And you can’t really talk about “rotation roles” without talking about innings pitched. Bumgarner may have pitched more like a no. 3 on a rate basis (3.53 DRA, +0.79 SD to league), but that many innings at that level of performance is incredibly valuable. And I want to reiterate that I am overfitting a normal distribution pattern here. I don’t think even the most strident among us would say there are only 14 no. 1 or no. 2 starters in baseball right now.

But as long as I am being a bad stathead, let’s cherry pick some names that “feel” like mid-rotation starters.

Anibal Sanchez

Like many on the list, Sanchez is somewhat a function of our endpoints. In 2013, he had a career year, finishing top five in the AL Cy Young voting. In 2015, he was a well-below-average starter. And in 2014, he was…Anibal Sanchez…ish. Over the balance of his career he feels the most like a mid-rotation starter to me. cFIP and DRA like him less than more traditional metrics, but by ERA+ for example, he was about ten percent better than league average most seasons, while not being particularly durable. That sounds like a low 3/high 4 to me.

Our lists only go back to 2007, but Baseball America had Sanchez as the 40th best prospect in baseball before the 2006 season,

Alex Cobb

Cobb pitched very well in 2013-2014, before Tommy John cost him all of 2015. This is another case where our metrics like him a little less than ERA+. But even using DRA, he was worth five wins in just a bit over 300 innings. That might be a bit over the bar here actually. Cobb never made a national list, and his one mention on a team list came before the 2012 season where he checked in at no. 19 with the remark “He has more control than stuff, but his plus changeup allows him to compete at the upper levels.” Call this one an over-performing backend profile.

Carlos Carrasco

Like Cobb, Carrasco might be above our mid-rotation bar on a rate level. But he also has struggled to stay on the mound, throwing just 360 innings over our 2013-15 window. Carrasco the prospect I remember well, specifically because of the kerfuffle he caused among BP readers when Kevin Goldstein gave him a five-star rating despite the perfect world projection being: “A good third starter, perhaps a bit more.” It took a winding and unorthodox path to get there, but score one for KG.

Hisashi Iwakuma

Hiroki Kuroda

Hyun-Jin Ryu

All three of these pitchers fit nicely for me in a qualitative way, but also have to be thrown out from our unscientific survey. None really qualified as prospects given their long professional track record in Asia before coming to MLB (at least they didn’t get written up by jamokes like me), but in the interest of an honest accounting, none of the three were really seen as slam dunk top-of-the-rotation arms when they emigrated.

Collin McHugh

Is Collin McHugh a no. 3 starter could be the secret concern of the piece. He pitched 26 terrible innings in 2013 for the Mets and Rockies, part of a DFA swap in between stops. He emerged with the Astros in 2014 after adding velocity and suddenly throwing his slider a ton, and pitched like a no 2. In 2015 a bit of the velocity gain and a few of the strikeouts disappeared, but he was still a league-average-innings-muncher. Depending on where you set the edge of the frame, it’s a very different film.

As a prospect, McHugh was the classic squint-and-you-can-see-it back end type. Four-pitch mix, nothing better than average, always a bit old for the level, command and pitchability guy. Definitely an overachiever.

Michael Wacha

Tyson Ross

Wacha and Ross are both closer to top-of-the-rotation arms, right?

Well, since his sterling 2013 debut that had fans of other NL Central teams cursing Cardinals devil magic once again, Wacha has been “just” above-average and only threw 288 innings across 2014 and 2015. But man, he feels like a number two.

As for Wacha as a prospect, well, take it away Jason Parks and team: “Wacha can still fit nicely as a third starter, and he could reach that ceiling very quickly.[x]

Ross has been more durable than Wacha since becoming a full-time starter. His raw numbers have been better too, but the Petco park effect makes him look like the most like the consistently good, but not spectacular, mid-rotation arm we are seeking.

KG agreed, referring to him as a “good third starter” when he clocked in at the back end of A’s top ten lists in 2010 and 2011.

Shelby Miller

The Diamondbacks won’t be thrilled to learn Miller is on this list, but he hasn’t quite yet lived up to the billing we gave him as the 16th-best prospect in baseball before the 2013 season. 2015 may have been a step in that direction though, and he’s only 25.

Jorge de la Rosa

De La Rosa is another that falls into the category of “pretty good pitcher, not enough innings.” And that is true even when you widen the scope past 2013-15.

Francisco Liriano

Liriano went from Rule 5 pickup to budding ace to injured ace to, you guessed it, “pretty good pitcher, not enough innings.”

We’re at the mercy of endpoints and overfitting and what not, but as you may have figured out by now, I am not overly concerned with scientific rigor. These lists of arms may not convince you that we public evaluators are that much better or worse at projecting no. 3 types, but I hope maybe it illuminated the same thing for you that it did for me. I am not entirely sure what a major-league no. 3 is. I spoke of qualitative traits above, and I think that is the “aye, there’s the rub” moment. Michael Wacha and Carlos Carrasco and Collin McHugh aren’t really no. 3s. They are sometimes no. 5s and sometime no. 1s. And of course, every pitcher is. It’s just Voros’ law applied to pitchers. We know that when Clayton Kershaw gets blown up, it’s an outlier, same as when Chris Heston throws a no-hitter.

But you don’t get to a 3.70 ERA in a particularly straight line. The top-of-the rotation and back-end guys are defined ex post facto. To harken back to KG one more time, he would often say on Up and In that you can only call an ace an ace after multiple years in the majors of high performance and heavy workload. The mid-rotation arm defies that, because he is never actually a mid-rotation arm for a discernible period of time. Over our three-year sample, that 3.70 ERA may come from a partial season where the pitcher is a no. 2 for 100 innings and then hurt, a year struggling as a back-end type, and a year as a league-average innings muncher.

The game here is all about comps. You can comp Julio Urias or Steven Matz to Clayton Kershaw (you shouldn’t, but you can). You can comp a Double-A lefty with a change and some pitchability to Tommy Milone. But who exactly are you comping Braden Shipley to? Calling it a failure of imagination is accurate perhaps, but not precise. The mid-rotation arm you want to comp him to doesn’t actually exist. You don’t know that you don’t know.

A case study, Robert Gsellman

On our 2016 Mets Top Ten, Gsellman clocked in at no. 9, with a likely role 45 (no. 5 starter). I would have also given him an OFP 55 (low no. 3/high no. 4). The mid-rotation mad libs fit well here, as his changeup lagged behind the fastball and curve, and there wasn’t quite enough command to support a sit 88-92 sinker. So he would be a command/change refinement type.

Development is not linear though, and it is rarely as simple as that. The Mets right-hander added a slider this Spring[xi] while in his first major-league camp. And on Opening Day he reportedly touched 98 and was sitting significantly higher than last season. I missed him by a day in Binghamton, but did confirm the velocity bump.

I finally saw him last weekend, and although he only touched 94, he was more consistently 91-93 without sacrificing much in the way of his 2015 command or movement. The slider flashed plus, as did the curve. And he got New Hampshire’s Sunday lineup to swing over his sinker. The change still lags behind, his clear fourth pitch now. And it still isn’t a huge fastball. I saw enough to bump the profile a half-grade on both OFP and likely, which puts him very comfortably in our mid-rotation profile and makes him the type of arm that could sneak onto the back end of some national lists.

So what don’t we know that we don’t know?

The slider is a very new pitch, and Gsellman is learning it in an organization that has seen frequent, huge player-development gains when pitching prospects get to the majors. And oh, a lot of that has to do with pitchers suddenly having 70- or 80-grade sliders develop quickly under the tutelage of Dan Warthen. Now, we can’t project that for everybody,[xii] but if this mid-rotation projection were to suddenly look light, the slider developing well past my future grade for Gsellman’s[xiii] would be a likely culprit.

Our unscientific survey expects him to miss in the other direction though, and I am still betting on grade jumps for three secondary pitches. And if you start doing that willy-nilly, well you can turn a lot of Double-A arms into mid-rotation starters. The fastball is still very hittable when it is up in the zone, and the Double-A lefties could spit on the change more often than not. Maybe a bit more effort in the delivery gets him to 92-95, and the slider gets good enough that he’s a setup guy. Or the package doesn’t quite come together and he is Brad Peacock.

But this is all about making a call.

Sticking your neck out on a guy and throwing one grade on him.

A statement: This is what I think he is.

So tl;dr, I am writing up Robert Gsellman as a mid-rotation starter. 60% of the time, it works every time[xiv].

Robert Gsellman

Born: 07/18/1993 (Age: 22)
Bats: Right Throws: Right
Height: 6' 4" Weight: 200
Easy tempo, easy arm action. Some wind-up. Three-quarters slot. Athletic on the mound and repeats well, but will add a bit of effort to reach top velocity. 1.3-1.4 out of the stretch and maintains velocity there.
Evaluator Jeffrey Paternostro
Report Date 05/04/2016
Affiliate Binghamton Mets (AA, Mets)
Dates Seen 5/1/2016
OFP/Risk 60/Medium
Realistic 50: no. 4 starter
MLB ETA 2017
Video No
Pitch Type Present Grade Future Grade Sitting Velocity Peak Velocity Report
Fastball 55 60 89-93 94 Heavy sinking fastball, worm-burner. At 89-91, it gets late downward bite that can get swings-and-misses. At 91-93 it loses some plane but still gets some armside run. Struggled at times with commanding it armside, but also got squeezed on that side of the plate. Can work both sides of the plate down in the zone with it when he is at his best.
Slider 45 55 83-85 87 Has only been throwing it for a few months, but it will already flash plus. Shape can vary. The best ones were 83-85 with late, hard tilt. At higher velocity, looks more like a Warthen slider, with more depth than run. Can start it in the zone consistently, but struggles to spot it to the outside corner for strikes. Could be convinced to go future 6 with another look in better weather given how fast it has developed already.
Curveball 45 55 76-80 81 Gsellman has remarked about how developing a slider has helped his curveball, but the results are mixed. Slider and curve bled together early in his outing, became more distinct pitches later. The curve gets flatter and slurvy in the low 80s, but shows more true 11-5 break in the upper 70s. Can bury or spot. Would flash plus to the backfoot against lefties, but he falls in love with trying to backdoor it for strikes at time. Command isn't all the way there, but potential above-average offering.
Changeup 30 40 82-84 84 Clear fourth pitch, though he threw it a lot to a lefty-heavy New Hampshire lineup. Flashes average with good sink, but struggles to get it to the outer half of the plate, will lose it out of his hand and miss armside. Show-me pitch at present, but enough feel to project a useable, if below-average major league offering.

Gsellman is your prototypical, durable mid-rotation pitching prospect. He offers a full four-pitch mix with a potential plus fastball (due to the movement and command) two breaking balls that project as above-average. The change-up needs further refinement, and he will never have a huge margin for error given the fastball velocity, but the addition of the slider in 2016 has given him a potential major league out pitch and bumped the profile a half grade. Could also fall back as a sinker/slider setup type if the profile stagnates, but I like him as a starter.

[i] Which to some readers may serve as a threat on par with Grover’s grave warning that there is a monster at the end of the book.

[ii] In the bottom of an old fashioned glass, muddle one demerara sugar cube with a few shakes of Hella Bitters Orange bitters (enough to coat the cube). Next, vigorously stir in a splash of boiling water to create a quick simple. Then, add 4 oz of Buffalo Trace bourbon. Stir again (less vigorously) to dissolve the sugar further. Now add a large ice cube and stir a bit more. Finally, garnish with a maraschino cherry and an orange slice, or, if you are feeling fancy, three luxardo cherries on a skewer. I also have a fondness for the Dale Degroff version with the muddled fruit, nothing wrong with boozy punch.

[iii] Montero never made a BP top 101, but prior to the 2014 season, clocked in at 68 and 85 on the Baseball America and lists respectively.

[iv] I have a couple theories why: 1. The lack of a swing-and-miss pitch meant more foul balls and more nibbling against major league hitters. 2. He got a lot of called strikes armside where the catcher set up just off the plate. You can fool a Double-A ump hitting the glove there, but you don’t get those calls as a MLB rookie.

[v] This is the most interpretive one I suppose, but also KG sort of nailed it?

[vi] It was pointed out to me recently (and amusingly) that we’ve thrown an 80 OFP on five eyewitness reports, and all five are Lucas Giolito.

[vii] Alternate take: “Dude, you’re getting a Dell(on) Gee?” No? Doesn’t work? Yeah, that is why it’s in the endnotes.

[viii] Or, put another way: There are way more players a couple weird breaks away from a month of major league per diems than we commonly acknowledge.

[ix] I’ll put the answer down here in case you wanted to guess…it’s Ryan Vogelsong.

[x] From the 2013 Annual, where Wacha clocked in as the 56th best prospect in baseball.

[xi] #shockednotshocked

[xii] It definitely didn’t happen for Jon Niese, to name one.

[xiii] As it did for Familia, and deGrom, and Harvey, and then I think I finally learned my lesson with Syndergaard and Matz

[xiv] All apologies to Dr. Walsh and Professor Cook for not referencing Rashomon here or something.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
I love the Dunning-Kruger reference. Well-done.
Great article! The offseason Braden Shipley madlib captured it for me. I think the Colin McHugh example illustrated an interesting related idea. If you have but one season of brilliance before failing to adapt, do you average as a mid-rotation arm? I'd still call him a BORP.

Then again, given how few #1s there are, how many #2s are asked to pitch as #1s, and thus 3s as 2s, 4s as 3s, and so forth. There is no such thing as a mid rotation arm.
This is an interesting article. But I've been puzzling over a slightly different question for a while: "Why do we even care what number we assign to a starting pitcher, outside of for the purpose of a playoff series?"

It's not like your #3 always matches up with the other team's #3. The order of the rotation -- like the batting order -- is a highly over-considered and over-rated part of building a winning team. Over 162 games, isn't having 5 starters who give you a chance to win each day more important than having a CLEAR #1 through #5?
Of course you are correct. It would only make a difference if the #1 starter made 40 starts and each subsequent starter made several fewer at each step in the rotation. Let us say 35-30-25-20 and a few spot starts then a Kershaw becomes far more valuable. As long as each spot in the rotation makes approximately 32 starts a year it is the overall quality of the entire rotation that determines the strength of the staff. When the Mets can send Steven Matz out every fifth day as their "4th starter" then you have something special.
Yeah, on the prospect side it is more about comps, or putting guys in buckets. Your #3 projection guy may end up the best pitcher on your staff (Hi Brewers) for a couple years. Or you might be Steven Matz, and have a 70 OFP, but be the fourth best starter in your rotation. It is descriptive rather than prescriptive.

The interesting thing to me was the idea that "mid-rotation arms" as we think about them are more slippery than the class of arms on either side.
This is, of course, how rotations used to work. For example, the 1963 Dodgers went 42-40-34-23-12-8-3-1 in GS. Those days are long gone, though.
Nice catch. This is a perfect example of the value of using top of the rotation pitchers in the most effective way possible, even though their arms may fall off prematurely.
Is it coincidence that Gsellman looks like a face-swap of Harvey and DeGrom?
"They're cloning them in a lab" is as good an explanation as any for the Mets recent player development success with pitchers.
I still think Shipley can be a no. 2. This is his what, 4th year pitching? His curveball has proved dramatically. Fastball command is still a bit fringy, but he's a great athlete and I think that'll come. If so, his changeup really is a plus plus pitch.

Just my opinion. But the flashes are there.
Seems like the Padres had a lock on mediocre 2012 pitching prospects.
They got them for a reliever, and Kelly was showing some progress till his unfortunate injury. Erlin actually was looking like he belonged earlier this year but now needs TJ.

Totally disagree on Ross being mid rotation type (but I know that's a different argument)
I don't have a problem with the Padres acquiring these players. In some ways it's a dubious distinction, they had 3 pitchers from the 2012 top 101. That's impressive.

It is interesting that these players have not become solid regular contributors. Makes me wonder what kind of success rate the Padres have compared to other teams. Of course this gets hairy when considering the amount of talent in the system, injuries, etc. but it could be a fun deep dive.
Where does Quintana fit on this list?
For what it's worth, social scientists based the 20-80 scale on a normal distribution with mean 50 and standard deviation 10, but once Branch Rickey started applying it to baseball, it lost all connection to such distributions and became merely a convenient language for scouting discussions (and later assessments of anything Baseball Twitter enjoys rating). It is certainly not the case that, to pick an example, the mean of the population is 50 speed.

I don't think there's a firm definition of a middle-of-the-rotation starter, but #61-#90 on a pref list of all the SPs in baseball seems like a reasonable start. Retrospectively, the #61-#90 DRA-WARP leaders from 2015 were worth 1.2-1.8 wins above replacement, with most of them within 10 points of average cFIP.
@myshkin - super interesting tidbit about Branch Rickey.

The way you defined a number 3 is pretty much exactly how I was looking at it a couple weeks ago. I took the average of all active pitcher WARs and RA9WARS over the last 3 years and put them in a list of 150 pitchers. The top 30 were the aces, 31 - 60 were #2s, etc. Here's how they broke down as groups per 32 starts:

#1: 206 IP, 2.93 ERA, 3.14 SIERA, 205 K, 19.0 K-BB%, 4.9 WAR combo
#2: 194 IP, 3.47 ERA, 3.70 SIERA, 175 K, 14.4 K-BB%, 3.4 WAR combo
#3: 188 IP, 3.78 ERA, 3.96 SIERA, 153 K, 12.3 K-BB%, 2.6 WAR combo
#4: 194 IP, 3.92 ERA, 4.05 SIERA, 152 K, 11.4 K-BB%, 2.1 WAR combo
#5: 183 IP, 4.23 ERA, 4.12 SIERA, 144 K, 10.6 K-BB%, 1.6 WAR combo

Some guys in the #3 starter range are Homer Bailey, Justin Verlander, Carlos Rodon, James Shields, Mike Fiers, and Julio Teheran. Seems like a simple enough, and accurate enough, barometer.
How does this piece get written without once mentioning Noah Syndergaard?